Optimisation of a ground anchor head for NDT monitoring

Peer reviewed

'Bayesian Optimisation for Nurse Scheduling'

A Bayesian optimisation algorithm for a nurse scheduling problem is presented, which involves choosing a suitable scheduling rule from a set for each nurse's assignment. When a human scheduler works, he normally builds a schedule systematically following a set of rules. After much practice, the scheduler gradually masters the knowledge of which solution parts go well with others. He can identify good parts and is aware of the solution quality even if the scheduling process is not yet completed, thus having the ability to finish a schedule by using flexible, rather than fixed, rules. In this paper, we design a more human-like scheduling algorithm, by using a Bayesian optimisation algorithm to implement explicit learning from past solutions. A nurse scheduling problem from a UK hospital is used for testing. Unlike our previous work that used Genetic Algorithms to implement implicit learning [1], the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The Bayesian optimisation algorithm is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, new rule strings have been obtained. Sets of rule strings are generated in this way, some of which will replace previous strings based on fitness. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. For clarity, consider the following toy example of scheduling five nurses with two rules (1: random allocation, 2: allocate nurse to low-cost shifts). In the beginning of the search, the probabilities of choosing rule 1 or 2 for each nurse is equal, i.e. 50%. After a few iterations, due to the selection pressure and reinforcement learning, we experience two solution pathways: Because pure low-cost or random allocation produces low quality solutions, either rule 1 is used for the first 2-3 nurses and rule 2 on remainder or vice versa. In essence, Bayesian network learns 'use rule 2 after 2-3x using rule 1' or vice versa. It should be noted that for our and most other scheduling problems, the structure of the network model is known and all variables are fully observed. In this case, the goal of learning is to find the rule values that maximize the likelihood of the training data. Thus, learning can amount to 'counting' in the case of multinomial distributions. For our problem, we use our rules: Random, Cheapest Cost, Best Cover and Balance of Cost and Cover. In more detail, the steps of our Bayesian optimisation algorithm for nurse scheduling are: 1. Set t = 0, and generate an initial population P(0) at random; 2. Use roulette-wheel selection to choose a set of promising rule strings S(t) from P(t); 3. Compute conditional probabilities of each node according to this set of promising solutions; 4. Assign each nurse using roulette-wheel selection based on the rules' conditional probabilities. A set of new rule strings O(t) will be generated in this way; 5. Create a new population P(t+1) by replacing some rule strings from P(t) with O(t), and set t = t+1; 6. If the termination conditions are not met (we use 2000 generations), go to step 2. Computational results from 52 real data instances demonstrate the success of this approach. They also suggest that the learning mechanism in the proposed approach might be suitable for other scheduling problems. Another direction for further research is to see if there is a good constructing sequence for individual data instances, given a fixed nurse scheduling order. If so, the good patterns could be recognized and then extracted as new domain knowledge. Thus, by using this extracted knowledge, we can assign specific rules to the corresponding nurses beforehand, and only schedule the remaining nurses with all available rules, making it possible to reduce the solution space. Acknowledgements The work was funded by the UK Government's major funding agency, Engineering and Physical Sciences Research Council (EPSRC), under grand GR/R92899/01. References [1] Aickelin U, "An Indirect Genetic Algorithm for Set Covering Problems", Journal of the Operational Research Society, 53(10): 1118-1126,

'Bayesian Optimisation for Nurse Scheduling'

A Bayesian optimisation algorithm for a nurse scheduling problem is presented, which involves choosing a suitable scheduling rule from a set for each nurse's assignment. When a human scheduler works, he normally builds a schedule systematically following a set of rules. After much practice, the scheduler gradually masters the knowledge of which solution parts go well with others. He can identify good parts and is aware of the solution quality even if the scheduling process is not yet completed, thus having the ability to finish a schedule by using flexible, rather than fixed, rules. In this paper, we design a more human-like scheduling algorithm, by using a Bayesian optimisation algorithm to implement explicit learning from past solutions. A nurse scheduling problem from a UK hospital is used for testing. Unlike our previous work that used Genetic Algorithms to implement implicit learning [1], the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The Bayesian optimisation algorithm is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, new rule strings have been obtained. Sets of rule strings are generated in this way, some of which will replace previous strings based on fitness. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. For clarity, consider the following toy example of scheduling five nurses with two rules (1: random allocation, 2: allocate nurse to low-cost shifts). In the beginning of the search, the probabilities of choosing rule 1 or 2 for each nurse is equal, i.e. 50%. After a few iterations, due to the selection pressure and reinforcement learning, we experience two solution pathways: Because pure low-cost or random allocation produces low quality solutions, either rule 1 is used for the first 2-3 nurses and rule 2 on remainder or vice versa. In essence, Bayesian network learns 'use rule 2 after 2-3x using rule 1' or vice versa. It should be noted that for our and most other scheduling problems, the structure of the network model is known and all variables are fully observed. In this case, the goal of learning is to find the rule values that maximize the likelihood of the training data. Thus, learning can amount to 'counting' in the case of multinomial distributions. For our problem, we use our rules: Random, Cheapest Cost, Best Cover and Balance of Cost and Cover. In more detail, the steps of our Bayesian optimisation algorithm for nurse scheduling are: 1. Set t = 0, and generate an initial population P(0) at random; 2. Use roulette-wheel selection to choose a set of promising rule strings S(t) from P(t); 3. Compute conditional probabilities of each node according to this set of promising solutions; 4. Assign each nurse using roulette-wheel selection based on the rules' conditional probabilities. A set of new rule strings O(t) will be generated in this way; 5. Create a new population P(t+1) by replacing some rule strings from P(t) with O(t), and set t = t+1; 6. If the termination conditions are not met (we use 2000 generations), go to step 2. Computational results from 52 real data instances demonstrate the success of this approach. They also suggest that the learning mechanism in the proposed approach might be suitable for other scheduling problems. Another direction for further research is to see if there is a good constructing sequence for individual data instances, given a fixed nurse scheduling order. If so, the good patterns could be recognized and then extracted as new domain knowledge. Thus, by using this extracted knowledge, we can assign specific rules to the corresponding nurses beforehand, and only schedule the remaining nurses with all available rules, making it possible to reduce the solution space. Acknowledgements The work was funded by the UK Government's major funding agency, Engineering and Physical Sciences Research Council (EPSRC), under grand GR/R92899/01. References [1] Aickelin U, "An Indirect Genetic Algorithm for Set Covering Problems", Journal of the Operational Research Society, 53(10): 1118-1126,

The contribution of building research to the independence and quality of life of older people

Many nations are experiencing rapid rises in the life expectancy of their citizens. The implications of this major demographic shift are considerable offering opportunities as well as challenges to reconsider how people should spend their later years. A key task is enhancing the quality of life of older people through enabling them to continue to live independently even though illness, accident or frailty may have severely reduced their physical and sensory abilities and, possibly, mental health. Yet the needs of older people and disabled people have been largely ignored in the design of everyday consumer products, the home, transport systems and the built environment in general. Whilst the need for designers, engineers and technologists to provide products, environments and systems which are inclusive of all members of society is widely accepted, there is little understanding of how this can be achieved. In 1998 the UK Engineering and Physical Sciences Research Council established its EQUAL Initiative. This has encouraged design, engineering and technology researchers in universities to join with their colleagues from the social, medical and health sciences to investigate a wide range of issues experienced by older and disabled people and to propose solutions. Their research, which directly involves older and disabled people and, for example, social housing providers, social services departments, charities, engineering and architectural consultants, and transport firms, has been extremely successful. In a very short time it has influenced government policy on housing, long-term care, and building standards, and findings have been taken up by architects, designers, health-care professionals and bodies which represent older and disabled people.

Direct mass measurements of (68)Se and (80)Y

The masses of neutron-deficient nuclides near the N=Z line with A=64-80 have been determined using a direct time-of-flight technique which employed a cyclotron as a high-resolution spectrometer. The measured atomic masses for (68)Se and (80)Y were 67.9421(3) u and 79.9344(2) u, respectively. The new values agree with the 2003 Atomic Mass Evaluation. The result for (68)Se confirms that this nucleus is a waiting point of the rp-process, and that for (80)Y resolves the conflict between earlier measurements. Using the present results and the 2003 Atomic Mass Evaluation compilation, the empirical interaction between the last proton and the last neutron in N=Z nuclei has been revisited and extended.

Endogeneous matching in university-industry collaboration: Theory and empirical evidence from the UK

We develop a two-sided matching model to analyze collaboration between heterogeneousacademics and firms. We predict a positive assortative matching in terms of both scientificability and affinity for type of research, but negative assortative in terms of ability on one sideand affinity in the other. In addition, the most able and most applied academics and the mostable and most basic firms shall collaborate rather than stay independent. Our predictionsreceive strong support from the analysis of the teams of academics and firms that proposeresearch projects to the UK's Engineering and Physical Sciences Research Council.

Annual CREMRU-JCI survey of corporate real estate practices in Europe and North America: 1993-2002

The annual survey of corporate real estate practices has been conducted by CREMRU since 1993 and in collaboration with Johnsons Controls Inc. since 1997. This year the survey forms the first stage of a broader research project: International Survey of Corporate Real Estate Practices: longitudinal study 1993-2002, being undertaken for the Innovative Construction Research Centre at the University of Reading, funded by the Engineering and Physical Sciences Research Council. The survey has been endorsed by CoreNet, the leading professional association concerned with corporate real estate, which opened it to a wider audience. This summary of the ten annual surveys focuses on the incidence of corporate real estate management (CREM) policies, functions and activities, as well as the assessment of knowledge or skills relevant to the CREM function in the future. Both are of vital interest to educational institutions concerned with this field, as well as the personnel and training functions within organisations concerned with better management of their property.

A socio-technical network approach to the adoption of low and zero carbon technologies in new housing

Housing in the UK accounts for 30.5% of all energy consumed and is responsible for 25% of all carbon emissions. The UK Government’s Code for Sustainable Homes requires all new homes to be zero carbon by 2016. The development and widespread diffusion of low and zero carbon (LZC) technologies is recognised as being a key solution for housing developers to deliver against this zero-carbon agenda. The innovation challenge to design and incorporate these technologies into housing developers’ standard design and production templates will usher in significant technical and commercial risks. In this paper we report early results from an ongoing Engineering and Physical Sciences Research Council project looking at the innovation logic and trajectory of LZC technologies in new housing. The principal theoretical lens for the research is the socio-technical network approach which considers actors’ interests and interpretative flexibilities of technologies and how they negotiate and reproduce ‘acting spaces’ to shape, in this case, the selection and adoption of LZC technologies. The initial findings are revealing the form and operation of the technology networks around new housing developments as being very complex, involving a range of actors and viewpoints that vary for each housing development.

Mutual Information Rate and Bounds for It

The amount of information exchanged per unit of time between two nodes in a dynamical network or between two data sets is a powerful concept for analysing complex systems. This quantity, known as the mutual information rate (MIR), is calculated from the mutual information, which is rigorously defined only for random systems. Moreover, the definition of mutual information is based on probabilities of significant events. This work offers a simple alternative way to calculate the MIR in dynamical (deterministic) networks or between two time series (not fully deterministic), and to calculate its upper and lower bounds without having to calculate probabilities, but rather in terms of well known and well defined quantities in dynamical systems. As possible applications of our bounds, we study the relationship between synchronisation and the exchange of information in a system of two coupled maps and in experimental networks of coupled oscillators.

KDIGO-based acute kidney injury criteria operate differently in hospitals and the community—findings from a large population cohort

ACKNOWLEDGEMENTS We acknowledge the data management support of Grampian Data Safe Haven (DaSH) and the associated financial support of NHS Research Scotland, through NHS Grampian investment in the Grampian DaSH. S.S. is supported by a Clinical Research Training Fellowship from the Wellcome Trust (Ref 102729/Z/13/Z). We also acknowledge the support from The Farr Institute of Health Informatics Research. The Farr Institute is supported by a 10-funder consortium: Arthritis Research UK, the British Heart Foundation, Cancer Research UK, the Economic and Social Research Council, the Engineering and Physical Sciences Research Council, the Medical Research Council, the National Institute of Health Research, the National Institute for Social Care and Health Research (Welsh Assembly Government), the Chief Scientist Office (Scottish Government Health Directorates) and the Wellcome Trust (MRC Grant Nos: Scotland MR/K007017/1).

Absorption cross-sections of hole polarons in glassy and β-phase polyfluorene

Absorption induced by electrochemically injected holes is studied in poly-9,9-dioctylfluorene (PFO) films. Injected charges form positive polarons which are delocalised over four fluorene units in the glassy phase and about seven fluorene units in its β-phase. Polaron absorption cross-sections at the 640 nm peak are similar to the published values of chemically reduced oligofluorenes in solution. The absorption cross-section of polaron in the β-phase at 470 nm is about eight times smaller than the stimulated emission cross-section derived from published data. This indicates that β-phase-rich PFO is an attractive candidate for a light-emitting layer in double-heterostructure organic laser diodes.

Vortices in the extended Skyrme-Faddeev model

We construct analytical and numerical vortex solutions for an extended Skyrme-Faddeev model in a (3 + 1) dimensional Minkowski space-time. The extension is obtained by adding to the Lagrangian a quartic term, which is the square of the kinetic term, and a potential which breaks the SO(3) symmetry down to SO(2). The construction makes use of an ansatz, invariant under the joint action of the internal SO(2) and three commuting U(1) subgroups of the Poincare group, and which reduces the equations of motion to an ordinary differential equation for a profile function depending on the distance to the x(3) axis. The vortices have finite energy per unit length, and have waves propagating along them with the speed of light. The analytical vortices are obtained for a special choice of potentials, and the numerical ones are constructed using the successive over relaxation method for more general potentials. The spectrum of solutions is analyzed in detail, especially its dependence upon special combinations of coupling constants.

Helical buckling of Skyrme-Faddeev solitons

Solitons in the Skyrme-Faddeev model on R-2 x S-1 are shown to undergo buckling transitions as the circumference of the S-1 is varied. These results support a recent conjecture that solitons in this field theory are well-described by a much simpler model of elastic rods.

The role of lipids in mechanosensation

Acknowledgements: This work was supported by Wellcome Trust grants WT092552MA (J.H.N. and I.R.B.), Senior Investigator Award WT100209MA (J.H.N.), 093228 (T.K.S.) and 092970 (M.S.P.S.), and Biotechnology and Biological Sciences Research Council grants BB/I019855/1 (M.S.P.S.), BB/H017917/1 (J.H.N. and I.R.B.) and BB/J009784/1 (H.B.). We acknowledge the Diamond Light Source for beam time. I.R.B. is supported as a Leverhulme Emeritus Fellow. J.H.N. is supported as a Royal Society Wolfson Merit Award holder and as a 1000 Talent Scholar at Sichuan University. A.C.E.D. was supported by an Engineering and Physical Sciences Research Council Systems Biology Doctoral Training Centre student fellowship. We thank R. Phillips, A. Lee and S. Conway for helpful discussions.